This disclosure relates to glucose sensors and to methods of manufacture thereof.
The control of Type I Diabetes Mellitus is generally effected by the periodic injection of insulin to maintain blood glucose levels as close to normal as possible. The blood glucose level is monitored by means of a device that directly measures glucose from a blood sample. Insulin is injected in the appropriate quantities and at the appropriate intervals to correct imbalances in the blood glucose level. Careful control of blood glucose levels is mandatory for preventing the onset of complications such as retinopathy, nephropathy and neuropathy. Unfortunately in many cases, patients neglect to perform regular glucose monitoring and therefore suffer episodes of hyperglycemia or hypoglycemia, which may, in turn, lead to the complications listed above or to death.
Blood-glucose levels generally vary with activity or food intake and insulin is therefore administered by sub-cutaneous hypodermic injection to minimize variations in the blood glucose levels that generally occur with activity or food intake. Small externally worn pumps are also available to deliver insulin percutaneously, thereby replacing the tedious use of a hypodermic injection, but constant glucose monitoring is still an important component of control.
Attempts to develop a closed loop system for the control of glucose levels have led to the development of ever more sophisticated insulin pump systems. However, an accurate long lived implanted blood glucose level monitor that would provide the desired signal for a closed loop insulin pump control is not yet available. An implanted blood glucose level monitor hinges on the accuracy of measuring glycemic levels in diabetic patients, thereby imposing stringent requirements in the confidence level of the continuous monitoring technology. In recent years, three kinds of glucose sensors are being developed: non-invasive, minimally-invasive and invasive.
Non-invasive techniques acquire spectroscopic information through skin or from various body fluids/gases (i.e., saliva, tears, and breath) and attempt to correlate this with glucose concentration. Non-invasive techniques generally use explanted sensors. Minimally-invasive sensors measure glucose concentrations from fluids obtained from the interstitial tissue of the skin via microdialysis, iontophoresis, laser ablation, and silicon-based micro-needle technologies. Both non- and minimally-invasive methods use elaborate calibration schemes and have considerable subject-to-subject variability.
Invasive methods use implanted sensors. These are generally advantageous in that they exhibit smaller subject-to-subject variability. However they are associated with a number of other problems. In particular, inflammation associated with tissue injury and the continuous presence of a foreign object is exacerbated by implant size and the presence of leads or fluid-microcatheters protruding through the skin. This constitutes the main cause of sensor failure in vivo, along with sensor element decays due to long-term usage.
Tissue injury-based sensor bio-instability is considered to be a result of the in vivo environment since explanted sensors often function normally without giving rise to any problems. It is generally believed that inflammation initiated fibrosis, calcification, and protein fouling are the leading causes of in vivo sensor failure. Implantation trauma, lack of biocompatibility of sensor materials and the physical presence of the sensor in the tissue are responsible for such tissue responses. Negative tissue responses (such as, biofouling, inflammation causing fibrosis and calcification) inhibit analyte migration and hence sensor performance; long-term sensor stability; and in vivo sensor calibration. Fibrous encapsulation can deprive the sensor of adequate blood and analyte supply. This can be modeled by effectively changing the permeability constants of the membrane(s) that surrounds the sensing element.
The D-glucose (dextrose monohydrate) specificity of analyte-specific enzymes such as glucose oxidase (GOx), have helped propel Clark-type electrochemical detection as a major technological frontier in the development of implantable glucose sensors. The most commonly used glucose sensors are Clarke-type amperometric electrochemical sensors and are based on GOx-catalyzed oxidation of glucose with O2, shown in reaction (1). The principle of detection is based on the amperometric sensing of hydrogen peroxide (H2O2), formed by the oxidation of glucose. Under an applied potential of 0.7 V against a silver/silver chloride (Ag/AgCl) reference electrode, H2O2 is electrochemically oxidized according to reaction (2), and the current produced is related to the concentration of glucose in the system.

In testing methodology, the sensor is biased continuously at 0.7 V while the change in electrochemical response is measured, which in turn corresponds to the glucose levels. For the accurate performance of these sensors it is desirable that the amount of oxygen present within the sensor geometry must always be equal or higher than that of the glucose concentration. However, the dissolved oxygen concentration in ambient or in a biological fluid sample is significantly lesser than that of the glucose concentration, leading to an oxygen limiting reaction of the GOx enzyme. This results in a saturation of the electrochemically detected signal, making it impossible to determine higher levels of glucose in the blood. As a result of this saturation in the amperometric signal (defined as apparent Michael's constant Kmapp) any further increase in glucose concentration does not translate to adequate sensitivity.
This issue has been addressed by the use of diffusion limiting outer membranes. These membranes provide a greater impendence to the larger sized substrate (glucose) as opposed to the smaller sized co-substrate (O2). For this, semipermeable membranes based on NAFION®, polyurethane, cellulose acetate, epoxy resins, polyether-polyethersulfone copolymer membranes, and layer by layer (LBL) assembled polyelectrolytes and/or multivalent cations have been extensively investigated. However, the use of semipermeable membranes comes at the expense of decreased sensitivity and increased sensor response time. Furthermore, the accumulation of exogenous reagents within these membranes (i.e., calcification, biofouling, or the like) leads to sensor drifts and their eventual failure.
In another variation, an additional oxygen reservoir can be incorporated into the outer membrane by incorporating oxygen-absorbing zeolites. Similarly, oxygen reservoirs such as fluorocarbon based oxygen reservoirs, mineral oils and myoglobin can be incorporated into the glucose oxidase enzyme layer.
In another variation, second- and third-generation Clark type biosensors employ redox mediators and direct ‘wiring’ of enzymes to electrodes in an attempt to minimize the effect of O2. In the case of mediators, their toxicity and biocompatibility along with the possibility to leach out from the device to the surrounding tissue present a major problem. Direct wiring of enzymes to electrodes can minimize the oxygen limitation, although this modification adds unwanted complexities and higher expense.
These defects have been rectified by developing a polarographic technique for simultaneous measurement of oxygen and glucose. However, the low sensitivity of the electrode (in the polarographic technique) to oxygen and the involvement of oxygen in the oxidation of other interfering species (i.e., ascorbic acid (AA), acetaminophen (AP), uric acid (UA), and the like) render the method unsuitable for reliable operation. Independent determination of glucose and oxygen concentrations could in principle account for oxygen induced sensor interferences. A number of reports have attempted to account for these, although addition of other sensor element adds additional complexities with respect to sensor integration, testing and calibration.
As mentioned above, an impediment with Clark-type glucose sensors is the fact that a number of endogenous species, such as ascorbic acid (AA), acetaminophen (AP), uric acid (UA), and the like), also oxidize at the same potential as H2O2 (i.e. 0.6-0.7 V), which can add error to the electrochemical signal. High confidence sensors have to actively account for these species, and at present not many methodologies have been developed. For example, anionic charged membranes based on negatively charged polymers (e.g., NAFION®, polyester sulfonic acid, cellulose acetate, and the like) have shown to exclude interferences from anionic species like ascorbic acid, uric acid, and the like, based on the principle of charge repulsion. However, the large response time associated with these membranes hinders their usage. Another popular approach to eliminate interference signals from endogenous species has been the use of inner, ultra-thin, electropolymerized films between working electrode and enzyme layer. These films have been shown to exert partial screening from interference agents to first generation analyte sensors. However, these electropolymerized films only minimize signal from endogenous species, and eliminating such interference has not been realized. Moreover, these membranes do not possess long term stability, and their interference eliminating property decreases shortly due to swelling of the polymer.
In another approach, secondary enzymes (for example ascorbate oxidase which converts ascorbic acid to dehydroascorbate and water) have been incorporated in the outer membrane of the sensor to eliminate the particular species from reaching the electrode surface and contributing to amperometric current. These secondary enzymes do however use oxygen as a co-substrate and could eventually deplete the sensors from the oxygen that is used for the operation of the primary enzyme (i.e. GOx). In yet another approach, independent determination of these interferences using secondary working electrodes have improved sensor reliability, although, once again, the addition of another sensor adds additional complexities involving sensor integration, testing and failure.
Another major problem associated with these implantable sensors is the changes in the electrocatalytic activity of the working electrodes as well as the in the permeability of the outer membranes after implantation in the body. While the former is a result of product adsorption on the surface of the working electrode, the latter is a result of unwanted accumulation of exogenous reagents within these membranes (i.e., calcification, biofouling, and the like). Such factors lead to decrease in sensitivity, drifts, and to their eventual failure. Moreover, passivation of working electrodes and inhibition of its electro-catalytic activity as result of continuous biasing also leads to saturation in sensor response. To this end, higher applied potentials, double pulsed amperometry or pulsed amperometric detection have been the common strategies to renew the surface of the working electrode even though such techniques are complex to be applied for miniaturized sensors and implantable sensors with miniaturized driving electronics. To date there is no reported methodology to account for such in vivo induced sensor drifts and the ability to internally calibrate the sensor against these variations is paramount for long-term sensor operation.